Below are the first 10 and last 10 pages of uncorrected machine-read text (when available) of this chapter, followed by the top 30 algorithmically extracted key phrases from the chapter as a whole.
Intended to provide our own search engines and external engines with highly rich, chapter-representative searchable text on the opening pages of each chapter. Because it is UNCORRECTED material, please consider the following text as a useful but insufficient proxy for the authoritative book pages.
Do not use for reproduction, copying, pasting, or reading; exclusively for search engines.
OCR for page 82
82 CHAPTER 7 Bicycle LOS Model 7.1 Development LOS grade response was taken from the overall dataset for model validation. The balance of the data, 80% of the total Two basic forms were considered for the bicycle LOS for dataset, was used for model development. arterials model. The first was an aggregate model using the SPSS 14.0 was used to conduct Pearson correlation analysis outputs from existing segment and intersection LOS models on the extensive array of geometric and operational variables. to determine the arterial LOS. The other was an agglomerate Subsequently, we selected the following relevant variables for model considering the independent characteristics of the additional testing: roadway environment to calculate an arterial LOS for bicy- clists directly. Both forms were preliminarily evaluated dur- · Segment LOS--The bicycle LOS for roadway segments ing model development. (see below). The aggregate model was chosen for refinement for several · Intersection LOS--The bicycle LOS for signalized inter- reasons. The stepwise approach to an aggregate model is use- sections (see below). ful because it allows the practitioner to address concerns at · Conflicts per mile--The total conflicts per mile represent individual intersections or along specific segments. The ag- the motor vehicle conflicts resulting from motorists turn- gregate model also retains all the terms found both intuitively ing across the bicycle facility at unsignalized locations. and mathematically to be significant to bicyclists riding along · Size of the city in which the data collection took place-- a roadway. The agglomerate model would not retain all the The Metropolitan Statistical Area (MSA) population was terms as significant. Consequently, we focused on the aggre- used to represent the size of each city. gate model in model development efforts. We considered various functional techniques for model At the panel's request, the MSA variable was dropped development, including linear regression and ordered probit. from further consideration. Other variables were dropped We performed linear regression modeling because it is more from further consideration because of their poor correla- intuitive than probit modeling in practice and non-modelers tion with the dependent variable or because of their colin- better understand the sensitivity of the regression model. earity with more strongly correlated variables. After testing These reasons are particularly important in that these mod- numerous combinations of variables and variable transfor- els are most frequently used: the development or analysis of mations, we determined the aggregate model using two specific design options or in the development of bicycle facil- constituent sub-models would be the most theoretically ity community master plans with presentations to interested valid. citizens and public officials. To ensure the validity of the re- sults of the linear regression modeling results, we evaluated the ordered probit model form as well. The results of both the 7.2 Recommended Bicycle linear regression and ordered probit modeling efforts are de- LOS Model scribed below. Before starting correlations analysis and modeling, we cre- The recommended bicycle LOS model is a weighted ated two data subsets from the overall dataset. The total combination of the bicyclists' experiences at intersections and dataset was sorted by city and LOS grade responses. A ran- on street segments in between the intersections. Two models of dom sampling of 20% of the data representing each city and the same form were evaluated, but with different parameters: